Re: [Development] Enabling private features for separate modules

[Alexander Volkov]

19.06.2017 20:06, Oswald Buddenhagen пишет:

On Mon, Jun 19, 2017 at 01:52:59PM +0300, Alexander Volkov wrote:

19.06.2017 12:53, Oswald Buddenhagen пишет:

On Mon, Jun 19, 2017 at 12:52:12PM +0300, Alexander Volkov wrote:

Hi,

I build qtwebengine as a separate module.
What is the right way to enable system-ffmpeg feature?


you use

qmake [usual qmake args] -- [configure args]


$ qmake -- -system-ffmpeg
***Unknown option -system-ffmpeg


you need to specify a file/directory, if only ".".


Yes, it works this way, thanks.
Is there a reason for such behavior or is it a bug?
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] Enabling private features for separate modules

[Oswald Buddenhagen]

On Mon, Jun 19, 2017 at 01:52:59PM +0300, Alexander Volkov wrote:
> 19.06.2017 12:53, Oswald Buddenhagen пишет:
> > On Mon, Jun 19, 2017 at 12:52:12PM +0300, Alexander Volkov wrote:
> >> Hi,
> >>
> >> I build qtwebengine as a separate module.
> >> What is the right way to enable system-ffmpeg feature?
> >>
> > you use
> >
> >qmake [usual qmake args] -- [configure args]
> >
> 
> $ qmake -- -system-ffmpeg
> ***Unknown option -system-ffmpeg
>
you need to specify a file/directory, if only ".".

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] What changed in the RHEL 6.6 machines that cause tst_QUdpSocket::linkLocalIPv6 to fail?

[Thiago Macieira]

On Monday, 19 June 2017 03:48:42 PDT Morten Sørvig wrote:
>  - Can we self-test the network interface and QSKIP if it’s not in working
> order?

That's the test itself: it tries to send a packet back to itself and see if it 
arrives.

>  - We can blacklist on CI only:
> https://codereview.qt-project.org/#/c/197692/ (modifying that patch to
> blacklist all Linux CI if needed.)

I don't believe in blacklisting if we can find and fix the root cause. That's 
what I did for the "utun" interface on Darwin machines, when a blacklist was 
proposed. But I haven't been able to figure that out for RHEL yet.

-- 
Thiago Macieira - thiago.macieira (AT) intel.com
  Software Architect - Intel Open Source Technology Center

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] Enabling private features for separate modules

[Allan Sandfeld Jensen]

On Montag, 19. Juni 2017 11:52:12 CEST Alexander Volkov wrote:
> Hi,
> 
> I build qtwebengine as a separate module.
> What is the right way to enable system-ffmpeg feature?
> It's not auto-detected:
>   "system-ffmpeg": {
>   "label": "ffmpeg",
>   "autoDetect": false,
>   "condition": "libs.ffmpeg && features.system-opus &&
> features.system-webp",
>   "output": [ "privateFeature" ]
>   }
> 
If all else fail you can always edit the generate qtwebengine-config.pri file 
and move the feature from disabled features to enabled features.

' Allan

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] Flaky auto tests on macOS 10.12

[Morten Sørvig]

> On 19 Jun 2017, at 13:09, Tony Sarajärvi  wrote:
> 
> It is the intention that we do this. But we're constantly running, not beside 
> our horse, but after it. We have just begun activating ourselves to discuss 
> about 5.10 platforms and we have FF coming up in two weeks (counting out July 
> here).
> It took months to blacklist all the cases for 10.12 alone. So yes, I agree 
> that how we do this is not good, but at some point we find it good enough. I 
> ran qt5 builds through twice without failures and deemed it fit. Before that 
> I kept blacklisting everything I saw (which actually isn't good either).
> 
> Dev branch is actually the place where they should be put yes. Unfortunately 
> we are too late for that. We need 10.12 in 5.9, and for it to still be in dev 
> branch, we should have put it in January. We were still working with our 5.8 
> platforms at that time. As I said...the horse is running away from us 


The process could be (using the current branches and os versions as examples):

- Add 10.12 to dev.
- Stabilize/fix/blacklist (backport changes to 5.9)
- Qt is now tested on 10.12. Also the 5.9 codebase, since it is similar to dev.

- Then add 10.12 to 5.9
- With less disruption since Qt already has been stabilized.

Morten
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] Enabling private features for separate modules

[Alexander Volkov]

19.06.2017 13:59, Dominik Holland пишет:

$ qmake -- -system-ffmpeg
***Unknown option -system-ffmpeg

I think it needs to be:

qmake -- --enable-system-ffmpeg

if you want to add your own options you can also add a "commandline"
section to configure.json
It also doesn't work. Somehow this way of passing extra arguments with 
'--' doesn't work in release builds.

I tested it in my developer build and it works.
Also I found that it was introduced by 
https://codereview.qt-project.org/#/c/159056/

and there is another way of passing extra arguments with QMAKE_EXTRA_ARGS:
qmake QMAKE_EXTRA_ARGS+="-system-ffmpeg" works.
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] Flaky auto tests on macOS 10.12

[Tony Sarajärvi]

It is the intention that we do this. But we're constantly running, not beside 
our horse, but after it. We have just begun activating ourselves to discuss 
about 5.10 platforms and we have FF coming up in two weeks (counting out July 
here).
It took months to blacklist all the cases for 10.12 alone. So yes, I agree that 
how we do this is not good, but at some point we find it good enough. I ran qt5 
builds through twice without failures and deemed it fit. Before that I kept 
blacklisting everything I saw (which actually isn't good either).

Dev branch is actually the place where they should be put yes. Unfortunately we 
are too late for that. We need 10.12 in 5.9, and for it to still be in dev 
branch, we should have put it in January. We were still working with our 5.8 
platforms at that time. As I said...the horse is running away from us 

-Tony

-Original Message-
From: Development 
[mailto:development-bounces+tony.sarajarvi=qt...@qt-project.org] On Behalf Of 
Morten Sørvig
Sent: maanantai 19. kesäkuuta 2017 13.40
To: development@qt-project.org
Subject: Re: [Development] Flaky auto tests on macOS 10.12


> On 19 Jun 2017, at 08:34, J-P Nurmi  wrote:
> 
> 
> [*] The timing, during the 5.9.1 soft freeze week, was a bit unfortunate, but 
> this is a separate discussion.

Introducing a new configuration to the CI system should be considered a 
development task, subject to the usual branch policies. Can we establish that 
this happens in dev first, ideally before the feature freeze?

Morten

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] Enabling private features for separate modules

[Dominik Holland]

Am 06/19/2017 um 12:52 PM schrieb Alexander Volkov:

> 19.06.2017 12:53, Oswald Buddenhagen пишет:
>> On Mon, Jun 19, 2017 at 12:52:12PM +0300, Alexander Volkov wrote:
>>> Hi,
>>>
>>> I build qtwebengine as a separate module.
>>> What is the right way to enable system-ffmpeg feature?
>>>
>> you use
>>
>>qmake [usual qmake args] -- [configure args]
>>
>
> $ qmake -- -system-ffmpeg
> ***Unknown option -system-ffmpeg

I think it needs to be:

qmake -- --enable-system-ffmpeg

if you want to add your own options you can also add a "commandline"
section to configure.json

Dominik
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] Enabling private features for separate modules

[Alexander Volkov]

19.06.2017 12:53, Oswald Buddenhagen пишет:

On Mon, Jun 19, 2017 at 12:52:12PM +0300, Alexander Volkov wrote:

Hi,

I build qtwebengine as a separate module.
What is the right way to enable system-ffmpeg feature?


you use

   qmake [usual qmake args] -- [configure args]



$ qmake -- -system-ffmpeg
***Unknown option -system-ffmpeg
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] What changed in the RHEL 6.6 machines that cause tst_QUdpSocket::linkLocalIPv6 to fail?

[Morten Sørvig]

> On 16 Jun 2017, at 23:21, Thiago Macieira  wrote:
> 
> On Friday, 16 June 2017 13:23:07 PDT Thiago Macieira wrote:
>> QDEBUG : tst_QUdpSocket::linkLocalIPv6(WithoutProxy)
>> QHostAddress("fe80::250:56ff:feab:4818%eth1")
>> FAIL!  : tst_QUdpSocket::linkLocalIPv6(WithoutProxy)
>> '(neutralReadSpy.count()
>>> 0)' returned FALSE. ()
>> 
>>Loc: [../tst_qudpsocket.cpp(1600)]
> 
> Emergency workaround: https://codereview.qt-project.org/197753 
> (compiles and works for me on bare metal, no test on a VM)
> 
> This commit basically disables the test entirely on virtual machines. If any 
> QtNetwork developers use virtual machines, please make sure you get 
> bare-metal 
> hardware for all three platforms. You won't be doing your job if you continue 
> with VMs.


This does seem a bit heavy-handed to me.

 - Can we self-test the network interface and QSKIP if it’s not in working 
order?

 - We can blacklist on CI only: https://codereview.qt-project.org/#/c/197692/
   (modifying that patch to blacklist all Linux CI if needed.)

Morten
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

[Development] Qt 5.9.1 initial change files available

[Jani Heikkinen]

Hi all

initial change files available, see 
https://codereview.qt-project.org/#/q/message:%22Add+changes%22+branch:5.9.1,n,z

Maintainers: Please do needed modifications & get approval as soon as possible 
(today/latest tomorrow)

br,

Jani Heikkinen
Release Manager

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] Enabling private features for separate modules

[Oswald Buddenhagen]

On Mon, Jun 19, 2017 at 12:52:12PM +0300, Alexander Volkov wrote:
> Hi,
> 
> I build qtwebengine as a separate module.
> What is the right way to enable system-ffmpeg feature?
>
you use

  qmake [usual qmake args] -- [configure args]

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

[Development] Enabling private features for separate modules

[Alexander Volkov]

Hi,

I build qtwebengine as a separate module.
What is the right way to enable system-ffmpeg feature?
It's not auto-detected:
 "system-ffmpeg": {
 "label": "ffmpeg",
 "autoDetect": false,
 "condition": "libs.ffmpeg && features.system-opus && 
features.system-webp",

 "output": [ "privateFeature" ]
 }

Cheers,
Alexander.
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] What changed in the RHEL 6.6 machines that cause tst_QUdpSocket::linkLocalIPv6 to fail?

[Simon Hausmann]

Hi,


I've checked that all iptables for IPv4 and IPv6 are empty.



Simon


From: Development  on 
behalf of Edward Welbourne 
Sent: Monday, June 19, 2017 11:36:19 AM
To: Thiago Macieira
Cc: development@qt-project.org
Subject: Re: [Development] What changed in the RHEL 6.6 machines that cause 
tst_QUdpSocket::linkLocalIPv6 to fail?

Thiago Macieira (16 June 2017 22:23)

> QDEBUG : tst_QUdpSocket::linkLocalIPv6(WithoutProxy) 
> QHostAddress("fe80::250:56ff:feab:4818%eth1")
>  FAIL!  : tst_QUdpSocket::linkLocalIPv6(WithoutProxy) 
> '(neutralReadSpy.count() > 0)' returned FALSE. ()
> Loc: [../tst_qudpsocket.cpp(1600)]

I note that this line ends in a comment:
QTRY_VERIFY(neutralReadSpy.count() > 0); //note may need to accept a 
firewall prompt

COIN folks: please check RHEL 6.6 hasn't auto-enabled some firewall !
ISTR that I've hit Red Hat on-by-default fire-walls that I had to turn
off when testing CFEngine in my last job ...

Eddy.
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] What changed in the RHEL 6.6 machines that cause tst_QUdpSocket::linkLocalIPv6 to fail?

[Edward Welbourne]

Thiago Macieira (16 June 2017 22:23)

> QDEBUG : tst_QUdpSocket::linkLocalIPv6(WithoutProxy) 
> QHostAddress("fe80::250:56ff:feab:4818%eth1")
>  FAIL!  : tst_QUdpSocket::linkLocalIPv6(WithoutProxy) 
> '(neutralReadSpy.count() > 0)' returned FALSE. ()
> Loc: [../tst_qudpsocket.cpp(1600)]

I note that this line ends in a comment:
QTRY_VERIFY(neutralReadSpy.count() > 0); //note may need to accept a 
firewall prompt

COIN folks: please check RHEL 6.6 hasn't auto-enabled some firewall !
ISTR that I've hit Red Hat on-by-default fire-walls that I had to turn
off when testing CFEngine in my last job ...

Eddy.
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

[Development] Flaky auto tests on macOS 10.12

[J-P Nurmi]

Hi all,

After a few successful qt5.git CI rounds, auto tests were enabled for macOS 
10.12 in the CI last week. [*] Even though many flaky auto tests had been 
already blacklisted on macOS 10.12 earlier 
(https://bugreports.qt.io/browse/QTBUG-58968), it became nearly impossible to 
integrate any qtbase patch due to random timer/eventloop/animation related auto 
test failures on macOS 10.12. As agreed earlier 
(http://lists.qt-project.org/pipermail/development/2017-February/028715.html), 
an action was taken to blacklist/skip several more auto tests on macOS 10.12:


Extend blacklisting of tst_QElapsedTimer::elapsed to cover macOS 10.12
- https://bugreports.qt.io/browse/QTBUG-58713
- https://codereview.qt-project.org/#/c/195625/

Blacklist tst_QGuiEventLoop::processEvents in macOS 10.12
- https://bugreports.qt.io/browse/QTBUG-61131
- https://codereview.qt-project.org/#/c/196087/

Blacklist flaky tst_QTimeLine tests on macOS 10.12
- https://bugreports.qt.io/browse/QTBUG-61037
- https://codereview.qt-project.org/#/c/195608/

Skip unreliable tst_QTimer::moveToThread() on macOS 10.12
- https://bugreports.qt.io/browse/QTBUG-59679
- https://codereview.qt-project.org/#/c/197769/

Blacklist flaky tst_QGuiEventLoop::testQuitLock() on macOS 10.12
- https://bugreports.qt.io/browse/QTBUG-61499
- https://codereview.qt-project.org/#/c/197791/

Blacklist tst_QParallelAnimationGroup::deleteChildrenWithRunningGroup()
- https://bugreports.qt.io/browse/QTBUG-61500
- https://codereview.qt-project.org/#/c/197792/


Quoting Lars:
"
Anybody who get’s one of those bug reports assigned please look at them 
immediately and try to see what’s going wrong. This is not always easy or 
straightforward, as they aren’t always reproducible outside the CI system. Here 
are a few pointers to help:

Have a look at https://wiki.qt.io/Writing_good_tests , and check that the test 
is following the rules there. Often tests fail more easily under high load, so 
this is something to check as well. If you’re working for The Qt Company, you 
have the additional option of creating a VM inside the CI system or running 
test builds of a pushed change in the CI system. If you’re not working for 
TQtC, ask someone who does and we can schedule a build of any patch you push to 
gerrit inside the CI on the platform of your choice (with results being 
reported back to the gerrit change).

If the test is flaky due to some external dependency (e.g. the network test 
server), you might want to file them as a subtask for getting a better test 
server in place and keep it blacklisted. In almost all other cases, you should 
try to fix the test (or the bug in Qt). If it's not possible to fix the test, 
think about how it could be rewritten. If the test is worthless (for example 
because it doesn’t test anything we haven’t covered in other ways), remove it.

In any case, please handle those bug reports quickly, as said above they will 
block the release until handled. Please don’t down-prioritize these bugs 
reports without very good reasons (and talking to the module maintainer).

I hope that as many people as possible will help in the effort. Fixing those 
flaky tests is quite some work right now, but in the longer term we will 
benefit us all when integrations go in more smoothly and we can more easily 
update qt5.git and get releases out.
"

[*] The timing, during the 5.9.1 soft freeze week, was a bit unfortunate, but 
this is a separate discussion.

--
J-P Nurmi

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

[Development] Qt 5.9.1 branching finalized

[Jani Heikkinen]

Hi all,

Final downmerge is now done and '5.9' is for '5.9.2' from now on. We will start 
doing Qt 5.9.1 "rc" now to see if there is still some issue to be fixed before 
official release. But the target is that current '5.9.1' is final content for 
the release so please not try to get in any nice-to-haves anymore. Change files 
are still missing and we will do those soon today. Let's get those in as soon 
as possible

br,
Jani

From: Development  on 
behalf of Jani Heikkinen 
Sent: Monday, June 12, 2017 2:35 PM
To: development@qt-project.org
Cc: releas...@qt-project.org
Subject: [Development] Qt 5.9.1 soft branching started

Hi all,

We have today started soft branching from '5.9' to '5.9.1'. That way we should 
be able to release Qt 5.9.1 still during June. So please use '5.9.1' for new 
changes targeted to Qt 5.9.1 release. We will do final downmerge from '5.9' to 
'5.9.1' this Friday (16.6) and after that '5.9' will be for Qt 5.9.2.

And please note: We are planning to do regular patch level releases for Qt 5.9 
series so no any nice-to-haves in '5.9.1', just fixes for really critical 
issues. All others must be done in '5.9' & can wait Qt 5.9.2 release. So 
'5.9.1' is just for getting Qt 5.9.1 quickly out... And at this time we really 
want to do Qt 5.9.1 on time meaning we won't delay the release because of 
issues which were already in Qt 5.9.0 release. And we will check case by case 
all new regressions if those could really delay the release or not: Qt 5.9.2 is 
planned to be released already on August.

br,
Jani Heikkinen
Release Manager
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

Re: [Development] First Qt 5.9.1 snapshot available

[Jani Heikkinen]

> -Original Message-
> From: André Hartmann [mailto:aha_1...@gmx.de]
> Sent: sunnuntai 18. kesäkuuta 2017 18.25
> To: Jani Heikkinen ; development@qt-project.org;
> inter...@qt-project.org
> Cc: releas...@qt-project.org
> Subject: Re: [Development] First Qt 5.9.1 snapshot available
> 
> Hi Jani,
> 
> I've got two questions regarding the 5.9.1 release:
> 
> 1. There have been no "changes file" templates provided so far. I guess this
> will lead to last-minute hectic if you want to release in June. It would be 
> good
> to have such templates as soon as the release branch is created.

To be honest, it should be maintainers who does these change files from the 
begining, not the release team. But because it seems to be really hard to be 
done we in the release team have done these initial ones lately. Doing these 
too early won't help at all because many changes might be missing and that's 
why we are planning to do these immediately after branching from '5.9' -> 
'5.9.1' is finalized. I know it will be tight but let's try to get it done 
really quickly at this time. But to be honest we need to find a better way to 
do this now when we are trying to do these reqular patch level releases...

> 
> 2. For 5.8.0, I saw that binary compatibility files were added. Did I miss
> something or is this discontinued for 5.9.0? How do you assure BC during the
> 5.9 patch releases?
> 

No, this is continuing. 5.9.0 bc files are still missing and 5.9.1 is still 
compared to 5.8.0. 5.9.0 bc files will be added as soon as possible

br,
Jani


> Best regards,
> André
> 
> Am 14.06.2017 um 08:00 schrieb Jani Heikkinen:
> > Hi all,
> >
> > We have first Qt 5.9.1 snapshot available via online installer. It is
> > a separate 5.9.1 node, not as an update to Qt 5.9.0. Instructions how
> > to install it here:
> > https://wiki.qt.io/How_to_get_snapshot_via_online_installer
> >
> > Content of the snapshot is based on https://codereview.qt-
> project.org/#/c/196614/, delta to Qt 5.9.0 as an attachment.
> >
> > RTA testing is ongoing and snapshot seems to be pretty much OK. So
> > please test the snapshot & report your effort via
> > https://wiki.qt.io/Qt59_release_testing
> >
> > And please note: We will do regular patch releases for 5.9 series and
> > so on we won't block the release because of old issues. Qt 5.9.1 is
> > released on June and 5.9.2 is targeted to be released on August so if
> > there is still bad issues open in 5.9.1 there is time to fix those for
> > soon coming Qt 5.9.2 (or even Qt 5.9.3 which should be out
> > ~September...)
> >
> > br,
> > Jani
> >
> >
> >
> >
> >
> > ___
> > Development mailing list
> > Development@qt-project.org
> > http://lists.qt-project.org/mailman/listinfo/development
> >

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development